Evaluating the Goal-Directedness of Large Language Models

Open in new window